Skip to main content

Table 5 Classification of models trained on black and white images

From: Activity landscape image analysis using convolutional neural networks

CollectionRFSVMCNNMetric
10.48 ± 0.010.44 ± 0.010.62 ± 0.02Accuracy
0.47 ± 0.010.45 ± 0.010.62 ± 0.02F1
0.21 ± 0.020.16 ± 0.010.43 ± 0.04MCC
20.46 ± 0.010.43 ± 0.010.61 ± 0.03Accuracy
0.46 ± 0.010.44 ± 0.010.61 ± 0.03F1
0.20 ± 0.020.15 ± 0.020.42 ± 0.04MCC
30.47 ± 0.010.46 ± 0.020.60 ± 0.02Accuracy
0.47 ± 0.010.46 ± 0.020.60 ± 0.02F1
0.20 ± 0.020.19 ± 0.030.41 ± 0.03MCC
40.45 ± 0.020.47 ± 0.030.54 ± 0.05Accuracy
0.45 ± 0.020.48 ± 0.030.54 ± 0.04F1
0.17 ± 0.030.21 ± 0.040.32 ± 0.07MCC
50.41 ± 0.030.39 ± 0.010.70 ± 0.05Accuracy
0.41 ± 0.030.39 ± 0.010.69 ± 0.04F1
0.12 ± 0.050.09 ± 0.020.54 ± 0.07MCC
60.52 ± 0.030.51 ± 0.020.69 ± 0.07Accuracy
0.52 ± 0.040.51 ± 0.030.69 ± 0.07F1
0.29 ± 0.050.26 ± 0.030.53 ± 0.10MCC
70.69 ± 0.020.68 ± 0.010.73 ± 0.02Accuracy
0.69 ± 0.020.68 ± 0.010.73 ± 0.02F1
0.53 ± 0.030.52 ± 0.020.59 ± 0.04MCC
  1. The table summarizes classification performance for color-coded 3D AL and Ref-AL images using RF, SVM, and CNN models trained on b/w images. All values reported are averages and standard deviations over 10 independent trials